Molecular Dynamics Simulations of (De)Hydrogenation Mechanisms of Solid-State Hydrogen Storage Materials

由于氢的高重量密度（7.7wt%），氢化镁（MgH 2）是固态储氢应用中研究最多的材料之一。然而，氢化和脱氢反应的动力学较差，阻碍了其在实际应用中的性能。在实验上，很难确定在氢化和脱氢反应期间在材料（即Mg/MgH 2）内部发生了什么。本项目的目的是开发准确的神经网络（NN）潜力的Mg-H系统。NN电位将从从头计算中训练。然后，这些NN势将用于大尺寸和长时间尺度的分子动力学模拟，以了解氢化和脱氢反应中涉及的关键步骤的原子机制。基于基本的理解，我们将为固态储氢应用设计具有改进动力学和良好热力学的新材料。

Because of the high gravimetric density of hydrogen (7.7 wt%), magnesium hydride (MgH2) is one of the most studied material for solid state hydrogen storage applications. However, the poor kinetics of the hydrogenation and dehydrogenation reactions have hindered its performance in practical applications. Experimentally, it is very difficult to characterise what is happening inside the materials, i.e. Mg/MgH2, during the hydrogenation and dehydrogenation reactions. This project aims to develop accurate neural networks (NN) potentials for the Mg-H system. The NN potentials will be trained from ab initio calculations. These NN potentials will then be used for large-size and long-timescale molecular dynamics simulations, in order to understand the atomistic mechanisms of key steps involved in the hydrogenation and dehydrogenation reactions. Based on the fundamental understanding, we will design new materials with improved kinetics and good thermodynamics for solid state hydrogen storage applications.